A training-free layer that turns an open LLM into a decision model. It reads typed choice, yes/no and score answers from one prefill, removes option-order bias with no labels and, from a few hundred labels, calibrates or fits a closed-form head. Weights stay untouched.
Native MLX FP16 conversion of Laya for Apple silicon. The same ModernBERT-large encoder, decision transformer and heads, running in MLX with no PyTorch; the checkpoint Laya Studio fine-tunes from.
Decides
choice, score, noul, classify, route
choice, score, noul
Architecture
anyjev
laya
Fine-tuned from
qwen/qwen3-8b
convai-innovations/laya
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
77.1%
—
Calibration error
0.034
—
Valid action rate
—
—
Median latency
—
—
p95 latency
—
—
Figures are from each model’s manifest; accuracy and latency are what the publishers report, on their own suites and hardware. Add a third model.
Questions
What is the difference between anyjev and laya-mlx?
anyjev is from Nokia Applied Research and laya-mlx from aac6fef. Both have open weights you can download and run. Both answer choice, score and noul questions. Only anyjev answers classify and route. laya-mlx is the smaller model, at 421M parameters to 8.0B.
Which is more accurate, anyjev or laya-mlx?
Only anyjev publishes an accuracy figure (77.1% on LocalLLaMA/typed-decisions (2,000 held-out decisions), L2 with 300 labels per question); laya-mlx does not, so there is no comparison to make without your own test.
Which is cheaper, anyjev or laya-mlx?
anyjev: Free (open weights). laya-mlx: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run anyjev or laya-mlx locally?
Yes, both: systemone pull nokia-applied-research/anyjev and systemone pull aac6fef/laya-mlx download the weights.
Evaluation suite
LocalLLaMA/typed-decisions (2,000 held-out decisions), L2 with 300 labels per question
